Studies on parameters affecting sinter strength and prediction through artificial neural network model

Abstract

Bed permeability, rate of reductant and productivity of blast furnace (BF) performance mainly depends on both iron bearing material but also carbonaceous material. Most of the BFs have the sinter being a major burden; hence, in JSW Steel Ltd, four sinter plants are operating to fulfill the four BF's requirement. For efficient BF operations, sinter plants are key units whose proper performance is vital to produce desired sinter strength. The tumbler index of the sinter is an important property of the sinter, and sinter strength depends on the raw material composition and machine parameters. For smooth sinter plants operation, changes to the operating conditions should be few and precise. To achieve this, a much better understanding of the mechanisms relating control inputs to a sinter production rate and quality needs to be established. In the present work, a neural network based model has been developed and trained relating sinter strength with a set of nine process variables, namely, basicity, Al₂O₃/SiO₂, MgO, MnO, FeO, moisture, coke breeze, burnthrough temperature and machine speed, to predict the tumbler index ( − 6.3 mm) of the sinter. The variables to which strength of the sinter was most sensitive were Al₂O₃/SiO₂, basicty, machine speed, and MgO, MnO and FeO. Tumbler index of the sinter was influenced by sinter porosity, which was itself determined by the firing temperature and green sinter mix carbon content. The predicted results were in good agreement with the actual data with < 3.5% error.